Skip to main content

Google Cloud Run

Platform · CaaS

Containerized services, jobs, functions, and worker pools on managed serverless infrastructure.

Containerized services, jobs, functions, and worker pools on managed serverless infrastructure.

Best for

Containerized APIs, web applications, background workers, scheduled jobs, functions, and GPU-backed inference that need managed autoscaling without operating Kubernetes.

Watch out for

Cloud Run combines several execution models with different billing, scaling, timeout, CPU, GPU, storage, and networking behavior. Cost and operational characteristics depend heavily on the selected configuration.

Workloads
  • AI Inference
  • APIs
  • Backend Services
  • Background Workers
  • Data Processing
  • Scheduled Jobs
  • Web Applications
Services
  • Compute
  • Functions
  • GPU Compute
Hosting model
  • Managed
Google Cloud Run executes containerized applications on managed infrastructure as request-driven services, run-to-completion jobs, and background worker pools; it also supports deploying functions. Services can autoscale to handle HTTP traffic and events, while jobs and worker pools have different invocation, scaling, and billing behavior. The capability list in this profile applies to Cloud Run services. Tagged revision previews can share external state rather than create an isolated backing stack.

Development & Execution

Supported platform capabilities, execution environments, development technologies, deployment models, integrations, and triggers.

Runtimes

  • Custom Runtime
  • Deno
  • .NET Runtime
  • JVM
  • Node.js

Languages

  • C#
  • Go
  • Java
  • JavaScript
  • Kotlin
  • PHP
  • Python
  • Ruby
  • Rust
  • TypeScript

Frameworks

  • Django
  • Express.js
  • FastAPI
  • Fastify
  • Laravel
  • Next.js
  • Ruby on Rails
  • Spring Boot
  • SvelteKit

Capabilities

  • Compute Autoscaling
  • Scale to Zero
  • Zero-Downtime Application Deployments
  • Deployment Rollbacks
  • Deployment Traffic Splitting
  • Preview Environments

Architecture

  • x64
  • ARM
  • GPU

GPU vendors

  • NVIDIA

Deployment & packaging

  • Docker / OCI
  • Serverless (Managed Runtime)

Integrations

  • Datadog
  • GitHub
  • GitLab
  • Terraform

Triggers

  • Scheduled Invocation
  • Database Events
  • Publish-Subscribe Topics
  • HTTP Requests
  • Queue Messages
  • Storage Events

Pricing

Published pricing structure, representative costs, and material billing factors.

Starting Price
$0.000024 Active CPU time
Primary Billing
Resource based Google Cloud Run Request-Based Usage
Free Tier
Recurring free allowances
Commitment
Request-based Cloud Run services are billed for active CPU, memory, and requests configured minimum instances also incur idle CPU and memory charges.

Google Cloud Run Request-Based Usage

Primary Pricing

Representative USD Tier 1 request-based service rates for Iowa (us-central1); Cloud Run prices vary by region and billing configuration.

ResourceBilling UnitPrice
Active CPU timeper vCPU-second$0.000024
Active memory timeper GiB-second$0.0000025
Requests Requests denied by IAM policy are not billedper 1 million requests$0.40
Idle minimum-instance CPU time Applies to configured minimum instances while idleper vCPU-second$0.0000025
Idle minimum-instance memory time Applies to configured minimum instances while idleper GiB-second$0.0000025

Free Tier

Recurring monthly Cloud Run allowances applied as a spending-based discount using us-central1 Tier 1 request-based pricing.

  • CPU time: 180,000 vCPU-seconds per month
  • Memory time: 360,000 GiB-seconds per month
  • Requests: 2,000,000 requests per month

Billing Commitment

Request-based Cloud Run services are billed for active CPU, memory, and requests; configured minimum instances also incur idle CPU and memory charges.

Infrastructure & Availability

Where workloads run, how deployment is scoped, and what underlying infrastructure the platform depends on.

Google Cloud Run deploys services, jobs, functions, and worker pools across 43 current regions. GPU, domain mapping, ephemeral disk, worker pool, and VPC connector availability can be narrower than the base regional footprint.

Execution

  • Single Region
  • Multi Region

Deployment scope

  • Regional

Infrastructure providers

  • Google Cloud Platform

Available regions (43)

LocationRegion ID
North America 11 regions
Iowa Cloud Run Tier 1 pricing region.us-central1
South Carolina Cloud Run Tier 1 pricing region.us-east1
Northern Virginia Cloud Run Tier 1 pricing region.us-east4
Columbus Cloud Run Tier 1 pricing region.us-east5
Dallas Cloud Run Tier 1 pricing region.us-south1
Oregon Cloud Run Tier 1 pricing region.us-west1
Montreal Cloud Run Tier 2 pricing region.northamerica-northeast1
Toronto Cloud Run Tier 2 pricing region.northamerica-northeast2
Los Angeles Cloud Run Tier 2 pricing region.us-west2
Salt Lake City Cloud Run Tier 2 pricing region.us-west3
Las Vegas Cloud Run Tier 2 pricing region.us-west4
Latin America 3 regions
Mexico Cloud Run Tier 1 pricing region.northamerica-south1
Sao Paulo Cloud Run Tier 2 pricing region.southamerica-east1
Santiago Cloud Run Tier 2 pricing region.southamerica-west1
Europe 13 regions
Finland Cloud Run Tier 1 pricing region.europe-north1
Stockholm Cloud Run Tier 1 pricing region.europe-north2
Madrid Cloud Run Tier 1 pricing region.europe-southwest1
Belgium Cloud Run Tier 1 pricing region.europe-west1
Netherlands Cloud Run Tier 1 pricing region.europe-west4
Milan Cloud Run Tier 1 pricing region.europe-west8
Paris Cloud Run Tier 1 pricing region.europe-west9
Warsaw Cloud Run Tier 2 pricing region.europe-central2
Berlin Cloud Run Tier 2 pricing region.europe-west10
Turin Cloud Run Tier 2 pricing region.europe-west12
London Cloud Run Tier 2 pricing region.europe-west2
Frankfurt Cloud Run Tier 2 pricing region.europe-west3
Zurich Cloud Run Tier 2 pricing region.europe-west6
Asia Pacific 12 regions
Taiwan Cloud Run Tier 1 pricing region.asia-east1
Tokyo Cloud Run Tier 1 pricing region.asia-northeast1
Osaka Cloud Run Tier 1 pricing region.asia-northeast2
Mumbai Cloud Run Tier 1 pricing region.asia-south1
Bangkok Cloud Run Tier 1 pricing region; Serverless VPC Access connectors are not supported.asia-southeast3
Hong Kong Cloud Run Tier 2 pricing region.asia-east2
Seoul Cloud Run Tier 2 pricing region.asia-northeast3
Singapore Cloud Run Tier 2 pricing region.asia-southeast1
Jakarta Cloud Run Tier 2 pricing region.asia-southeast2
Delhi Cloud Run Tier 2 pricing region.asia-south2
Sydney Cloud Run Tier 2 pricing region.australia-southeast1
Melbourne Cloud Run Tier 2 pricing region.australia-southeast2
Middle East & Africa 4 regions
Tel Aviv Cloud Run Tier 1 pricing region.me-west1
Johannesburg Cloud Run Tier 2 pricing region.africa-south1
Doha Cloud Run Tier 2 pricing region.me-central1
Dammam Cloud Run Tier 2 pricing region.me-central2

Availability reviewed · Availability source

Resource & Execution Limits

Memory

512 MiB default / 32 GiB max

Services; first-generation minimum can be 128 MiB

Concurrency

1,000

Per instance

Request timeout

60 minutes

Per request

Request payload

32 MiB

HTTP/1 request

Response payload

32 MiB

Non-streaming HTTP/1 response

Security & Compliance

Security Controls

Access Control
Resource-level IAM and authenticated invocation Cloud Run services/jobs
Encryption at Rest
Customer-managed encryption key for deployed container image Customer Cloud KMS configuration
Private Networking
VPC Service Controls perimeter and VPC ingress/egress Customer organization/network configuration
Audit Logging
Cloud Audit Logs Management coverage; data-access classes may require configuration
Code Signing
Binary Authorization deployment policy Customer-configured for services and jobs

Audits & Certifications

SOC Reports
SOC 2 Type II Cloud Run in scope
ISO Certifications
ISO/IEC 27017 and 27018 Cloud Run in scope
PCI DSS
Cloud Run in scope

Compliance

FedRAMP
FedRAMP High Authorized service and region/configuration scope
HIPAA
HIPAA-eligible use Google Cloud BAA and customer controls required

FAQ

Does Google Cloud Run support WebSockets?
Yes. Cloud Run supports WebSockets without additional platform configuration. WebSocket connections are treated as long-running HTTP requests, so they remain subject to the service’s configured request timeout and clients should reconnect when a connection ends.
Is Google Cloud Run free?
Yes, within Cloud Run recurring monthly allowances. The allowance includes 180,000 vCPU-seconds, 360,000 GiB-seconds of memory, and 2 million requests per month before paid usage.
How much memory can Google Cloud Run use?
Cloud Run services default to 512 MiB of memory and can be configured up to 32 GiB. The minimum is 128 MiB in the first-generation execution environment and 512 MiB in the second generation, with valid memory ranges also depending on the selected CPU allocation.

Explore